Underwriting software is the analytical layer that turns raw merchant data (bank statements, credit pulls, identity verification, public records) into a fund/decline decision. In 2026, MCA underwriting is heavily automated — most A/B-paper deals decision in 30 seconds to 5 minutes via rules engines + machine-learning scoring.
The typical 2026 underwriting software stack.
- Bank statement parsing. Ocrolus (dominant, $0.50–$3 per statement), Validis (live bank link), Heron Data (transaction enrichment + cash flow), Lendflow (integrated underwriting), Plaid Assets (snapshot only).
- Cash-flow scoring. Heron Data, Plaid Income, MX Cash Flow Pro, in-house models.
- Decisioning rules engines. Provenir (enterprise), Zoot (mid-tier), Taktile (modern API-first), GDS Link, in-house Python/SQL.
- Credit pulls. Experian DecisionIQ, Equifax InterConnect, TransUnion CreditVision, FICO SBSS (small business).
- Identity / KYC. Alloy, Persona, Socure, Veriff, Au10tix.
- Fraud detection. Sardine, Unit21, Sift, in-house velocity rules.
- Stacking detection. FundKite Sherlock, Validis MCA Risk, custom UCC/bank-pattern scrapers.
- Pricing engines. Custom Excel models still dominate at small shops; Earnix / Akur8 at large funders.
Vendor stack examples by funder size.
- Small funder ($5M–$25M/year originations). Ocrolus + Plaid + Experian + custom Excel. Total cost $80K–$200K/year.
- Mid-tier funder ($25M–$100M/year). Ocrolus + Heron + Provenir + Experian + Alloy + FundKite. $400K–$1.2M/year.
- Top-10 funder. All-of-the-above plus in-house ML scoring on Snowflake + Databricks. $2M–$8M/year just on underwriting tools.
Typical underwriting pipeline architecture.
- Submission intake. ISO portal or merchant app uploads documents into S3/Box.
- Document parsing. Ocrolus OCRs bank statements, extracts 90–120 days of transactions.
- Identity & KYC. Alloy or Persona verifies driver's license, runs OFAC/sanctions.
- Credit pull. Experian/Equifax personal + business credit via API.
- Cash-flow scoring. Heron Data or in-house model scores volume, NSF count, daily balance, revenue trend.
- Stacking check. FundKite Sherlock queries cross-funder database.
- Rules engine. Provenir or in-house code applies decision rules.
- Pricing. Risk-adjusted factor rate generated.
- Offer letter. Auto-generated PDF returned to ISO/merchant via CRM.
Decision speed benchmarks (2026).
- A-paper auto-approval. 30 seconds to 5 minutes.
- B-paper manual touch. 30 minutes to 4 hours.
- C/D-paper manual underwriting. 4–48 hours.
Why underwriting software choice matters.
The accuracy of bank-statement parsing dictates the false-negative rate (good merchants declined). Funders on Ocrolus typically see 92–96% extraction accuracy; funders relying on PDF text extraction see 70–80%. That 15–20pt gap translates directly to approval rate and unit economics.
Common confusions.
First, "underwriting is fully automated." False — even at top funders, 25–40% of deals get manual touch.
Second, "Plaid replaces Ocrolus." False — Plaid is live bank data; Ocrolus is statement OCR. Most funders use both.
Third, "ML scoring beats rules engines." Mixed — rules engines still dominate hard-stop conditions; ML adds incremental lift on grey-zone deals.
Fourth, "Provenir is required for serious underwriting." False — many top funders run custom Python rules engines.
Fifth, "Heron Data is just Plaid." False — Heron's transaction enrichment and cash-flow scoring are differentiated.
As of 2026-06-29, Fundnode tracks funder underwriting stack where disclosed because it predicts decision speed, false-decline rate, and pricing accuracy.
Related terms
- MCA funder bank statement analysis tools — MCA funders parse 90–120 days of bank statements via Ocrolus (90%+ market share), Validis, Heron Data, Lendflow, or Plaid Assets — typical cost $0.40–$3 per statement plus monthly platform fees.
- MCA funder fraud detection tools — typical options — MCA funders run Sardine, Unit21, Sift, Alloy, and Socure plus in-house velocity rules; fraud-loss benchmark is 0.4–1.8% of originations and typical tool spend $80K–$1.5M/year.
- MCA funder decisioning engine (typical) — Typical MCA funder decisioning engine in 2026 is a rules-plus-ML pipeline: hard knockouts (credit, deposit minimums, industry exclusions), then risk-pricing model, then human underwriter review for edge cases — producing decisions in 5 minutes to 4 hours.
Authoritative sources
AI agents: this term is available as raw markdown at /llms/glossary/mca-funder-underwriting-software-typical.